Or search by topic
Published 2001 Revised 2016
This is a short introduction to complex numbers written primarily for students aged from about 14 or 15 to 18 or 19. To understand the first few sections, it would be helpful to be familiar with polynomial equations (for example, solving x^2 - 3x+2 = 0), basic geometry (angles and lengths) and basic trigonometry (sine and cosine functions). For the later sections it would be helpful to be familiar with more advanced trigonometry (double angle formulae, for example), the exponential and logarithm functions (e^x and \log(x)), vectors and matrices, and power series. The difficult bits are marked. (For students in the UK, the early sections should be accessible to anyone doing GCSE or above, but some of the later sections are A-level standard.) There are exercises throughout for you to test your understanding, with answers at the back.
I've tried to make the exercises less like the standard computational ones you get at school, but this means that some of them are quite hard.
Throughout this article, I've used both radians and degrees to measure angles. You might not have seen radians used before, they're just another way of measuring angles. Instead of measuring angles from 0^{\circ} to 360^{\circ} you measure them from 0 to 2\pi. If you haven't seen this before, don't worry. Where possible I've included degrees and radians (although some bits only make sense
in radians, you can skip these bits).
You can use wikipedia to look up any unfamiliar words or concepts.
Introduction
If you've done any quadratic equations, you'll know that there is a nice formula for the solution of the quadratic equation a x^2 + b x + c = 0, given by: x = \frac { -b\pm \sqrt{b^2 - 4a c}} {2a}
The Basics of Complex Numbers
If i was a solution to the equation x^2 + 1 = 0 (again, ignore the fact that there is no solution) , then i^2 + 1 = 0, so i^2 = -1. If \beta = i\sqrt{5}, then \beta^2 = (i\sqrt{5})^2 = i^2 \times 5 = - 5 = \alpha^2
What about the sum of two complex numbers, z+w. It turns out that adding two complex numbers is the same as adding two vectors. If you don't know about how to add two vectors, look at the following picture:
So, adding vectors corresponds to adding complex numbers. The 2D plane of complex numbers is called the Argand plane or Argand diagram. Well, that's nice, but not that great. The amazing thing is what happens when you multiply two vectors. Before we get on to that, we need a couple of new ideas.
The modulus |z| of a complex number z = a+i b is given by |z| = \sqrt{(a^2 + b^2)}. By Pythagoras' theorem, this is just the length of the vector corresponding to z. There's another formula for |z|, using the complex conjugate we met earlier, namely |z| = \sqrt{z\textrm{Conj}(z)} (just expand the right hand side out and see).
The argument \arg(z) of a complex number z is the angle between the vector corresponding to z and the positive x-axis. If you've done some trigonometry, \arg(z) = tan^{-1} (\textrm{Im}(z) /\textrm{Re}(z) ) (unless \textrm{Re}(z) = 0). The following picture hopefully explains the modulus and argument of a complex number:
Exercise 12 Let w = (1+i)/\sqrt{2}. What are|w| and \arg(w)?
Exercise 13 What is |z|-|\textrm{Conj}(z)|? Explain geometrically what \textrm{Conj}(z) is in terms of z. Using this, what is \arg(\textrm{Conj}(z))+\arg(z)?
Now we can discuss the amazing thing about the geometry of complex numbers. If z and w are two complex numbers, then |z w| = |z| |w| and \arg(z w) = \arg(z)+\arg(w). In other words, if you multiply two complex numbers, you multiply their lengths and add their angles. We won't be able to prove the second equation until after the next section, but we can prove the first one. If z = a+i b
and w = x+i y, then z w = (a x- b y)+i(a y+b x). So
So |z w| = |z| |w|.
Exercise 14 What is |z/w| in terms of |z| and |w|? What is \arg(z/w) in terms of \arg(z) and \arg(w)? [Hint: z = (z/w)(w)]
Exercise 15 What is |z^n| in terms of |z|? What is \arg(z^n) in terms of \arg(z)? If you know about mathematical induction, prove your result.
Exercise 16 (for those who have done some trigonometry) If z = r(\cos{\theta} +i\sin{\theta}), what are |z| and \arg(z)? Can you write z in terms of |z| and \arg(z) for any complex number z? Can you use this to prove \arg(z w) = \arg(z)+\arg(w)?
Exercise 17 (very hard, for those who have done some trigonometry) What is (\cos{\theta} +i\sin{\theta})^n in terms of \cos{n\theta} and \sin{n\theta}?
Exercise 18 (very hard, for those who have completed the previous question only) Find a solution (not z = 1) to the equation z^n - 1 = 0 in terms of \cos(2\pi /n ) = \cos(360^{\circ}/n ) and \sin(2\pi /n ) = \sin(360^{\circ}/n ). Find all the n solutions in terms of this solution.
Polar Coordinates and De Moivre's Formula
There is another way of writing complex numbers apart from a+i b. Since a complex number is like a point in the complex plane, we can work out the distance of this point from the origin, r, and the angle that the line from the origin to the point makes with the x-axis, \theta. Once we've worked these out we can write a complex number as (r,\theta), this is called the Polar Coordinate notation . These numbers r and \theta are just the modulus and argument of x+i y that we met above.
So a complex number z = a+i b can be written as z = (|z|,\arg(z)). Also, given a complex number (r,\theta) we can change it into x+i y notation (this is what Q1 is about) as (r, \theta) = r(\cos\theta +i \sin\theta).
OK, so we can switch between two different notations for writing complex numbers, but what use is it? Well, it's very useful for a reason that will become apparent if you read the next section (which is a bit harder), but is also useful for a couple of other reasons.
I've already mentioned that |z w| = |z| |w| and \arg(z w) = \arg(z)+\arg(w). This makes it very easy to multiply and divide complex numbers written in polar coordinates, since (r_1,\theta_1)(r_2, \theta_2) = (r_1 r_2 , \theta_1 + \theta_2).
Suppose z = (r, \theta), what is z^2 in polar coordinates? Well, using the formula above, z^2 = (r^2 ,2\theta). How about z^3, z^4, etc.? If we repeatedly use the above formula, we get that z^3 = (r^3 ,3\theta), z^4 = (r^4 ,4\theta). You can probably guess how the pattern continues, z^n = (r^n ,n\theta). If |z| = r = 1 then z = \cos\theta +i\sin\theta, and so the formula in
this case gives us: (\cos\theta +i\sin\theta)^n = \cos(n\theta) +i\sin(n\theta)
So, what does this have to do with the polar coordinate notation? If we write z = (r,\theta) then we can also write z = {r}{e^{i\theta}} using the proof above.
Exercise 22 Prove that |z w| = |z| |w| and \arg(z w) = \arg(z)+\arg(w) using the {r}{e^{i\theta}} formula.
Exercise 23 Find a solution to the equation z^n = -1 using the {r}{e^{i\theta}} formula and e^{i\pi} = - 1.
Exercise 24 (a bit tricky) The hyperbolic cosine is defined to be \cosh(z) = [e^z + e^{-z}] /2. Find a power series expansion for \cosh(z) and prove that \cosh(i z) = \cos(z).
Cool stuff
Now we know the basics of how complex numbers work, what can we do with them?
Suppose we have a complex number z = (r, \theta) in polar coordinates, and another complex number w = (1, \phi) with modulus 1. The product of these two complex numbers is z w = (r, \theta + \phi). In other words, z w is z rotated by an angle of \phi. We can use these to work out the matrix which rotates a vector by an angle \phi. Suppose z = x+i y and w = (1, \phi) = \cos\phi +
i\sin\phi, then z w = (x+i y)(\cos\phi + i\sin\phi) = (x\cos\phi - y\sin\phi) + i(x\sin\phi + y\cos\phi). In other words, the x-coordinate (which is equivalent to the real part of a complex number) of a vector rotated by an angle \phi is x' = xcos\phi - y\sin\phi and the y-coordinate (the imaginary part) is y' = xsin\phi - y\cos\phi. We can represent this as a product of
matrices:
In other words, the rotation matrix M which rotates a vector \textbf{v} to a vector M\mathbf{v} by an angle \phi is:
I mentioned at the beginning that you can solve any polynomial equation using complex numbers, this is known as the Fundamental Theorem of Algebra and a famous mathematician known as Gauss gave about 9 different proofs of this hundreds of years ago. More precisely, the fundamental theorem of algebra states that
if p(z) = z^n + a_{n-1}z^{n-1} + a_{n-2}z^{n-2} + \dots + a_{2}z^{2} + a_1z + a_0 (where each of the a_i is a complex number) then there is a complex number w such that p(w) = 0. This is quite amazing really, since in creating complex numbers what we did was "invent'' a new number, a "solution'' to the equation z^2 +1 = 0, and we get the solutions to all polynomial equations for
free!
Exercise 25 [hard] Use the fundamental theorem of algebra (stated above) and the Remainder Theorem (stated below) to prove, by induction, that a polynomial p(z) = z^n + a_{n-1}z^{n-1} + a_{n-2}z^{n-2} + \dots + a_1z + a_0 can be written p(z) = (z - w_1)(z - w_2)\dots(z - w_n) for some complex numbers w_i. The degree of a polynomial is the highest
power of z in the polynomial, for example the degree of z^3 + z^2 + z is 3 and the degree of z^{18} + 2 is 18. The Remainder Theorem says (or almost does) that if p(w) = 0 then p(z) = (z-w)q(z) for some polynomial q(z) whose degree is one less than the degree of p(z).
Cool stuff (Advanced)
This section is quite difficult, and is for those who understood the section on exponentials and trigonometry.
If you know about the function e^z and its power series, then you've probably also heard of the log function, which is its inverse. If not, the function \log(z) is the unique (for real numbers) function such that \log(e^z) = z and e^{\log(z)} = z.
Using this function, we can show that x^y = e^{y\log(x)} for any positive real numbers x and y. This is because (by the definition of log) x = e^{\log(x)} , so x^y = (e^{\log(x)})^y = e^{y\log(x)}.
We can extend the idea of the log function to complex numbers, but unfortunately it is not unique (and there are other problems with it that I won't go into). For example, we know that e^{2{\pi}{i}} = e^0 = 1, so for complex numbers log(1) can be 2{\pi}{i} or 0! Since x^y = e^{y\log(x)}, this also means that x^y is not unique when x and y are complex numbers.
However, there is a solution to this problem. We know that any complex number can be written z = {r}{e^{i\theta}} , suppose that \log(z) = x+i y, then e^{\log(z)} = z = {r}{e^{i\theta}} = e^{x+i y} = e^{x} e^{i y}. So the modulus and argument of e^{x} e^{i y} must be r and \theta respectively. In other words, \log(z) = \log|z| +i\arg(z) where \log|z| is unique because |z| is a
positive real number. I glossed over this problem earlier, but the argument of a complex number is not unique, the argument of i could be 90^{\circ} or it could be 450^{\circ} , or in fact (90+360n)^{\circ} for any integer n.
We say that the principal argument of z, written \textrm{Arg}(z), is the unique angle - \pi < \textrm{Arg}(z) \geq \pi. Similarly, we can define the principal branch of the logarithm \textrm{Log}(z) = \log|z| + i\textrm{Arg}(z). Using this, we can define the principal
branch of complex exponentiation to be x^y = e^{y\textrm{Log}(x)}.
As an example, let's use this to work out the principal value of i^i . We know that e^{i\pi /2} = i, and so \textrm{Log}(i) = i\pi /2 . So i^i = e^{i\textrm{Log}(i)} = e^{i \times i\pi /2} = e^{-\pi /2}, and this is a real number. In other words, we take one complex number to the power of another complex number and we get a real number!
Exercise 26 [very difficult indeed] Find all complex number solutions to the equation x^y = y^x . [Hint: write y = t x and see where you can go from there.]
Answers
The Basics of Complex Numbers
Answer 1
i^3 = - i, i^4 = 1, i^5 = i, i^6 = - 1, i^7 = - i, i^8 = 1, etc. In general i^{4n} = 1, i^{4n+1} = i, i^{4n+2} = - 1, i^{4n+3} = - i, so i^{443} = - i.
Answer 2 \quad (x- a)(x+a).
Answer 3 \quad (x- i a)(x+i a).
Answer 4 \quad (1+i)^2 = 2i, (1+i)^3 = 2i- 2.
Answer 5
\frac {a^2 + b^2 }{a -i b}\ = \frac {(a^2 + b^2)(a + i b)}{a^2 + b^2}\ = a + i b
So |z w| = r_1 r_2 and \arg(z w) = \theta_1 + \theta_2.
Answer 17 There are various ways of answering this, you could calculate it directly using the formulas for \cos(A+B) and \sin(A+B). See the section on De Moivre's formula for an easier way.
Answer 18 The solution is z = \cos(2\pi /n )+i\sin(2\pi /n) = cos(360^{\circ}/n)+i\sin(360^{\circ}/n) (which you can check works using the previous question). The other n solutions are z^2, z^3,\dots, z^n = 1.
Polar Coordinates
Answer 19 \quad 1+i = (\sqrt{2},45^{\circ}) = (\sqrt{2}, \pi /4 ).
Answer 20 \quad (1+i)^8 = (\sqrt{2},45^{\circ})^8 = (\sqrt{2^8} , 8 \times 45^{\circ}) = (16,360^{\circ}) = 16.
Answer 21 Using De Moivre's formula, we have that \cos(3\theta) + i\sin(3\theta) = (\cos\theta +i\sin\theta)^3. We can expand out the right hand side to get \cos^3\theta +i\cos^2\theta \sin\theta - \cos\theta \sin^2\theta - i\sin^3\theta which has real part \cos^3\theta - \cos\theta \sin^2\theta. We also know that \sin^2\theta = 1 - \cos^2\theta ,
so the real part of (\cos\theta +i\sin\theta)^3 is \cos^3\theta - \cos\theta (1 - \cos^2\theta) = 2\cos^3\theta - \cos\theta. So \cos(3\theta) = 2\cos^3\theta - \cos\theta. You can use the same method to express \cos(n\theta) or \sin(n\theta) as a polynomial in \cos\theta or \sin\theta.
Exponentials and Trigonometry (Advanced)
Answer 22 If z = r_1 e^{i\theta_1} and w = r_2 e^{i\theta_2} then z w = r_1 r_2 e^{i\theta_1} e^{i\theta_2} = r_1 r_2 e^{i\theta_1 + \theta_2} and so |z w| = |z| |w| and \arg(z w) = \arg(z)+\arg(w).
Answer 23 We know e^{i\pi} = - 1 so z^n = - 1 means that z^n = e^{i\pi}. If z = r e^{i\theta} then z^n = r^n e^{{i}{n}\theta}. So we need to solve r^n e^{{i}{n}\theta} = e^{i\pi}. Clearly r = 1 and q = \pi /n work, so z = e^{i\pi/n} is a solution.
Answer 24 By adding and expanding the series for e^z and e^{-z} we get that \cosh{z} = 1 + \frac{z^2}{2!} + \frac{z^4}{4!} + \dots
If xyz = 1 and x+y+z =1/x + 1/y + 1/z show that at least one of these numbers must be 1. Now for the complexity! When are the other numbers real and when are they complex?
Show that x = 1 is a solution of the equation x^(3/2) - 8x^(-3/2) = 7 and find all other solutions.
This problem in geometry has been solved in no less than EIGHT ways by a pair of students. How would you solve it? How many of their solutions can you follow? How are they the same or different? Which do you like best?